Parametric Trajectory Distillation for Few-Step Video Generation
We introduce Parametric Trajectory Distillation (PTD), which lets the student parameterize teacher trajectory segments as polynomials and learn from teacher guidance along its own predicted path.
Key points
- Video diffusion and flow models require many sequential evaluations, making generation computationally expensive.
- Few-step distillation reduces this cost but poses a capacity allocation problem: a student must match the teacher's iterative generation with far less sequential computation.
- PTD is designed to let the learned curvature adapt to the backbone's predictive capacity, preserving motion and diversity.
- On the 33B audio-video MiniMax-H3, LoRA-trained PTD significantly improves diversity and naturalness over the state-of-the-art LightX2V Turbo.
Sources (1)
- [1]Parametric Trajectory Distillation for Few-Step Video GenerationarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 08:38 AM
We introduce Parametric Trajectory Distillation (PTD), which lets the student parameterize teacher trajectory segments as polynomials and learn from teacher guidance along its own predicted path.
Video diffusion and flow models require many sequential evaluations, making generation computationally expensive.
Extractive summary: sentences quoted from the sources.